Tonic-Clonic and Clonic Seizure Detection through Non-Contact Affordable Means

Abstract

Epilepsy is the condition of recurrent, unprovoked seizures and occurs when a person is predisposed to seizures because of a chronic pathological state (e.g., brain tumor) or a genetic susceptibility. Approximately 50 million people suffer from epilepsy globally, making it one of the most common neurological diseases in the world. Currently, epileptic seizures are monitored and diagnosed using EEG combined with video recording but no known methods exist that are non-contact, affordable, and easy to access. Here we developed a system to detect clonic and tonic-clonic seizures based on their rhythmic natures. In the present program, key facial features are identified and tracked over time. Their locations and their movement over time are recorded; these data are then processed and a cyclic pattern corresponding to a seizure can be detected. This demonstrates an automated method to detect seizures in a non-contact way using cameras. We anticipate our developments to be a start for more advanced seizure detection algorithms that work under a broader range of conditions, such as varying camera viewpoints, detecting seizures without movement (e.g., heart rate variability), and by analyzing whole-body motion. This has significant implications and uses for the world at large, but especially for people in less developed countries where there is a significantly higher amount of deaths related to epilepsy and where people are more likely to not have the resources to utilize more complex methods of detection.